D&D 5E Dice Rolling for beginning ability scores...redux

[Cleon]

One of the things I liked about AD&D and BECMI was rolling ability scores. Was pondering recently how it could be adapted to 5e and "levelling" beginning characters with feats. e.g. mid range = one feat, lucky rolling = zero feats, unlucky rolling = two feats

So rating would be determining the overall "value" of the dice rolled, maybe according to this table - question is - what point spread = "normal", and where are the bars to one more or one less feat (or more tweaks)

Huh?

But the default in Fifth Edition is rolling ability scores randomly, using the old best 3 of 4d6, rolled six times then assigned to ability.

Giving them a bonus feat for low stats might be unbalancing for certain characters, particularly those with an excellent number in their prime stat and mediocre or low ones in others. Low ability scores shouldn't be worth a lot of "compensation" or the system's too open to min-maxing. Consider, say an Intelligence 18 Wizard with 3s in Strength and Charisma. Being a hideous weakling doesn't make their magic any less powerful.

The "negative ability value" you've proposed for ability scores below 8 seem too high to me.

Using the proposed numbers, a character with the ability array 3, 3, 14, 16, 16, 18 is a 26 point character. Is such a character really slightly weaker than a regular point buy character?

If I had a starting PC with overly low abilities in my game, if I wanted to make their life easier I'd probably give them something like bonus equipment rather than adding something integral to their character like an extra feat.

To balance high or low ability scores, how about doing something with Ability Advancement? Maybe PCs with low stats advance faster. Perhaps they get +1 to three stats or +2 to one stat and +1 to a second stat until their stats stop being "low", however that's defined.

Contrariwise, perhaps a PC with high stats either gets a poorer Ability Advancement (+1 to one stat instead of the +2 to one or +1 to two that's normal) or they're only able to apply their Advancement to their weaker stats, so a character with 8, 10, 12, 17, 17, 18 that hits 4th level can only apply its two points of Ability Advancement to it 8, 10 or 12 abilities, rather than putting +1 on each of the 17s and getting three 18s.

 

[fluffybunbunkittens]

a character with a +2 in all scores is actually worse that another with a total smaller bonus and uneven distribution

Definitely. DnD is a game that rewards specialists, go home if you're the second-best at something. You could handle it by just valuing +3 and +4 higher, it's not insurmountable. You just need to have a mid-step where you assign a value to the mods (and note that 17 is better than 16, but 11 isn't realistically different from 10).

 

[Cleon]

Ability Score Point Cost​

ScoreCost
3 = - 9 (?)
4 = -7 (?)
5 = -5 (?)
6 = -3
7 = -1
8 = 0
9 =1
10=2
11=3
12= 4
13= 5
14= 7
15 = 9
16 =11 (?)
17 = 13 (?)
18 = 15 (?)

For the high ability scores, an increase that raises the ability modifier (like, going from 15 to 16) might bump up the point cost more than one that doesn't (like 16 to 17) depending on how you're making Ability Advancement work.

With the current rule for Ability Advancement, odd high level scores are useful targets for the "increase two abilities by +1" rule.

If, however, you changed it so Ability Advancement gave you 4 more points to spend on abilities you can divide as you see fit, that'd work. Going from two 14s (cost 14) to two 15s (cost 18) or from one 13 (cost 5) to one 15 (cost 9) both increases the Ability Point Score by 4.

Hmm… this has possibilities.

Since Ability Advancement mostly happens once per 4 levels, you could change it so you get one AP of Ability Advancement per level, spreading out the increase. You could use it to raise mediocre stats quickly if you make under-8 points worth fractions. Maybe a 6 is –1 and a 5 is –½.

Characters with "good" ability bonuses could get fewer APs with their levels than those with "poor" ability bonuses. Somehow.

Maybe reset the Table so it starts at AP 0 for an ability score roll of 3 for the benefit of people who have trouble with negative numbers…

Perhaps something like this:

Ability Score	Ability Advancement Value	PHB Ability Score Point Cost
3	½	—
4	1	—
5	1½	—
6	2	—
7	2½	—
8	3½	0
9	4	1
10	5	2
11	6	3
12	7	4
13	8	5
14	9	6
15	10	7
16	12	—
17	14	—
18	16	—
19	18	—
20	20	—

…although I doubt I can be bothered fiddling with the idea further as I'm currently OK with standard random rolling.

 

[Casimir Liber]

Point buy is pretty comparable to rolling if you limit the range of possible numbers from a roll. The limit on the high number kind of counterbalances the fact that you're never going to have a number below an 8. Of course that kind of goes out the window with people that don't follow official rules and allow re-rolls or set a lowest number acceptable.

I prefer point buy because I don't want one PC to have significantly higher numbers than another PC. But that's a separate issue.

Yeah I know...but I just loved rolling dice for stats over the years

 

[Casimir Liber]

Contrariwise, perhaps a PC with high stats either gets a poorer Ability Advancement (+1 to one stat instead of the +2 to one or +1 to two that's normal) or they're only able to apply their Advancement to their weaker stats, so a character with 8, 10, 12, 17, 17, 18 that hits 4th level can only apply its two points of Ability Advancement to it 8, 10 or 12 abilities, rather than putting +1 on each of the 17s and getting three 18s.

I recall in AD&D, a Wish spell could be used to improve an ability score by 1 point up to 16, but then on was 1/10th of a point per wish (!)

 

[Cleon]

Definitely. DnD is a game that rewards specialists, go home if you're the second-best at something. You could handle it by just valuing +3 and +4 higher, it's not insurmountable. You just need to have a mid-step where you assign a value to the mods (and note that 17 is better than 16, but 11 isn't realistically different from 10).

Depends on edition.

In 3E and later, where you get a +1 higher modifier at each even ability score, a 17 is little different from a 16. It only let's you carry a bit more if its in Strength and makes creatures slightly more resistant to powers that lower their ability scores.

In earlier editions, there could be quite a difference in benefits.

There's still virtually nothing between 10 and 11 though!

If I were doing a Heartbreaker Game (i.e. a "what D&D would be like if I wrote it") I'd try to give the odd numbers a bit more love!

Maybe split the ability modifiers into two categories and have one go up on even scores and the other go up on odd scores. Like a 13 gives a +2 to Skill-type checks and +1 to Power-type or Resistance-type checks while a 14 gives +2 to both.

Or maybe have a modified version of the 4E approach, where you got +1 at each even ability score above 10 and +1 for each even level. Just add the ability score to the level and have +1 for each two points above 10. So a Strength 13 1st-level character has a +2 Str modifier as that adds up to 14.

 

[Cleon]

Yeah I know...but I just loved rolling dice for stats over the years

Me to!

To me, the big advantage is that random rolling can lead to quirkier and more interesting characters.

A point-buy character of a particular class can be a bit samey if there's only a few popular or optimal builds.

Plus you'll rarely get characters with high scores in abilities irrelevant to class features like a fighter with a high Intelligence or a Sorcerer with a high Strength in point buy, but it's not uncommon in randomly rolled characters.

 

[EzekielRaiden]

Okay, I couldn't wait until morning. If my late at night calculations in Excel are correct, then the variance of the ability bonus for one roll is 2.093555 (with an average of 0.873457). For six rolls, since the rolls are independent, we can just sum the variances, and the variance for the total bonus should be 12.56133, and the standard deviation would be 3.544196. So if you have +2 to +9 for your total bonus you are basically within one standard deviation of the norm.

But that's based on probability, which is not really the same as feat equivalency. So it would need to be more like what aco175 is saying. You could base it off the mean of +5 total bonus, but I would base it off who rolled best. But I would actually avoid something like this. I find first level feats to be awfully powerful.

Shouldn't the SD be that of the base roll (so, presumably, 4d6 drop lowest, which has SD 2.85 according to Anydice) times the square root of the number of dice rolled? (Technically, you add the SDs in quadrature, but when all of them are the same, this is equivalent to multiplying by the square root of the numbers so used.) SD is the square root of variance, but the former is what is actually useful for describing the range of a set of data

The center of the data for 4d6k3 is 12.24, and (thankfully) that does just multiply by six for the stat total. So we get 73.44 average, SD 2.85×√6=6.98. Within reasonable tolerance, that means overall stat totals between 59 and 88, a range of 29...which is exactly why things can be so problematic with character stats. A difference of 20 points is quite possible, and yet that translates to anywhere between a maximum of +12 higher total modifiers (e.g. {11, 11, 11, 11, 10, 10} vs {14, 14, 14, 14, 14, 14}) and a minimum of +7 (e.g. {12, 12, 10, 10, 10, 10} vs {15, 15, 15, 13, 13, 13}.) Of course, few things are going to result in such perfectly aligned results, so something in the middle, around +9, is reasonable as the potential spread between the lowest and highest rolling players.

This is where the rules for cutting off the bottom of the distribution come into play. 5e, like most games that propose rolling stats (including old school D&D, believe it or not!) includes rules for throwing out unacceptably weak results. If the highest stat is not at least 14, or the sum of modifiers is (IIRC) not positive, throw it out and start over. But that is in fact met by a slight rearragement of the examples I proposed: {14, 10, 10, 10, 10, 10}. So the minimum modifier gap remains unchanged. The maximum does fall, however, from +12 to +11. Which would seem to ideally position +9 as the "reasonable gap between best and worst rolls" amount.

If I were going to use this method, I would do the following.

1. Players roll their stats (six lots of 4d6k3) as normal and add any racial bonuses. Reroll any set that has no result of at least 14.
2. Players sum the total of their modifiers, and separately count the total number of odd scores they have.
3. If every player has a modifier total no more than 2 less than the highest player's total, do nothing--these scores are already good...or at least good enough.
4. If any player has a modifier total outside that range, they get free feats, one for every 2 points by which their total falls short.
5. If any player receives feats, they should also count their total number of odd ability scores. If they have more than two odd ability scores, they may (if they wish) also pick one from a list of flavorful "half" feats (e.g. Actor, Chef, Keen Mind.)

This method probably under-values feats (after all, you only get one for every 4-6 points by which the best rolled stats exceed your own, and get nothing at all for a difference of up to 6 points), but I figure the ability to select what you want makes the difference. I wish I didn't have to enforce a rigid list for point 5, but unfortunately, Elven Accuracy, one of the most powerful feats in 5e, is a half feat and must thus be accounted for.

With the racial bonuses, especially after Tasha's, having that guaranteed minimum highest stat of 14 means you always get at least one 16, so whatever your prime stat is, it's decent. And if you somehow managed to get a negative sum of modifiers, well, you'll probably be rolling in feats.

 

[EzekielRaiden]

Me to!

To me, the big advantage is that random rolling can lead to quirkier and more interesting characters.

A point-buy character of a particular class can be a bit samey if there's only a few popular or optimal builds.

Plus you'll rarely get characters with high scores in abilities irrelevant to class features like a fighter with a high Intelligence or a Sorcerer with a high Strength in point buy, but it's not uncommon in randomly rolled characters.

I find this is not nearly as common as people think, because it is roll and assign.

You only get actually unexpected stuff in one of two situations:
The player has rolled stupidly well and thus gets to "waste" good scores on irrelevant stats, or
The stat rolls were strict (e.g. Strength is the first roll) and thus the player cannot move the second(third/etc.) good roll to a more useful stat.

Because no matter what, even in ye olden dayse where people never rolled anything but 3d6 strict*, people would almost always pick a class that suited their ability scores. Everyone is "cookie cutter" in that sense. You won't find Fighters with garbage Str and Con and Dex, because that just won't be fun to play. You won't find Wizards with garbage Int, because that just won't succeed much. Note, this "you won't find" doesn't mean it's impossible, but it isn't impossible with point buy either...you'll just almost never see it either way, because people have the choice to play something effective vs ineffective, and effective will almost always win.

As long as people have the freedom to choose where to put their stats, almost every character will be cookie-cutter to some degree unless the player gets very lucky. As long as people have the freedom to choose their class, then even if rolls are strict, almost every character will be minimally cookie-cutter because almost every player will choose a class that fits their stats at least a little.

*This was never the default method for D&D, people just think it was.

 

[ichabod]

Shouldn't the SD be that of the base roll (so, presumably, 4d6 drop lowest, which has SD 2.85 according to Anydice) times the square root of the number of dice rolled? (Technically, you add the SDs in quadrature, but when all of them are the same, this is equivalent to multiplying by the square root of the numbers so used.) SD is the square root of variance, but the former is what is actually useful for describing the range of a set of data

I'm not calculating the SD of the ability scores, I'm calculating the SD of the ability bonuses. Other than that, yeah the math works out. But I calculated the variance and it was easier to multiply and then take the square root.